Research On Tag Recommendation For Online CQA Questions Based On Deep Learning

With the rapid development of Internet technology,people's access to information has become increasingly convenient.Various information sharing platforms have emerged.Online CQA(Community Question Answer)have emerged and attracted more and more users,which have gradually become an important platform for sharing information.A large number of new questions are generated every day.In order to effectively manage and organize these question data,the community classify questions using tags and provide one or more tags to each question.Tagging is different from the top-down hierarchical classification method of the general directory structure.The relationship among tags are parallel,which refines the classification with less cost.The online CQA effectively addresses the issue of question resource management by tagging.At present,the online CQA question tags is given by the user when the question is asked.Unfortunately,tagging process is distributed and uncoordinated due to users' understanding of their question,English skills and preferences.In order to solve this problem,some tag recommendation methods have been proposed to recommend a series of high-quality tags for users,which improves the efficiency and accuracy of question tagging.In this paper,tag recommendation models based on deep learning are proposed.The deep learning method is used to extract the semantic features of the questions to recommend tags for CQA questions.And the data enhancement method is used to improve the model performance.In addition,a combined model based on the basic deep learning models is proposed.The main work of this paper is as follows:(1)We regard the tag recommendation for online CQA questions as a content-based tag recommendation problem,and summarize the current research methods.Then we research relevant theories and techniques.(2)We select appropriate data objects,and extract the data.Then the data are preprocessed and word vector are trained to provide data support for the studied in this paper.(3)Three kinds of tag recommendation models based on deep learning are constructed.The influence of parameters on the model is studied.The effect of tag recommendation of deep learning models and traditional method are compared.(4)The data enhancement method is used to train the model and the combined model is proposed.The effects of them are studied through experiments.The experimental results in this paper show that the three tag recommendation models based on deep learning make better performance than the traditional recommendation methods,and the data enhancement method can further improve the performance of the model.The performance of the combined model is better than the three basic models.